POLYBIUS''PREVIOUS LANDSLIDE': PROOF THAT HANNIBAL'S INVASION ROUTE CROSSED THE COL DE LA TRAVERSETTE

Mediterranean Archaeology and Archaeometry, Vol. 14, No 2, pp. 77-94 Copyright © 2014 MAA

Printed in Greece. All rights reserved.

POLYBIUS’ ‘PREVIOUS LANDSLIDE’: PROOF THAT HANNIBAL’S INVASION ROUTE CROSSED THE COL

DE LA TRAVERSETTE

W.C. Mahaney1, C.C.R. Allen2, P. Pentlavalli2, Randy Dirszowsky3, Pierre Tri-cart4, Leslie Keiser5, Peeter Somelar6, Brian Kelleher7, B. Murphy7, P.J.M. Costa8,9,

S. Jordan7, S. O'Reilly7 and P. Juligj10

1Quaternary Surveys, 26 Thornhill Ave., Thornhill, Ontario, Canada, L4J 1J4 and Department of Geography, York University, North York, Ontario, Canada, M3J 1P3, 2Queen’s University Belfast, School of Biological Sciences, Belfast, N. Ireland, UK., 3Department of Earth Sciences, Laurentian University, Sudbury, Ontario, Canada, P3E 2C6, 4Institut des Sciences de la Terre, University of Grenoble, Observatoire des Sciences de l’Univers, 38041 Grenoble, France, 5Department of Geolog-ical Sciences, University of Oklahoma, Norman, Oklahoma, U.S., 73019-1009, 6Institute of Ecology and Earth Sciences, University of Tartu, Tartu, Estonia, 7School of Chemical Sciences, Dublin City University, Dublin, Ireland, 8Centro de Geologia da Universidade de Lisboa, Faculdade de Ciênci-as da Universidade de Lisboa, Edifício C6, Campo Grande, 1749-016 Lisboa, Portugal, 9Department of Geography, School of the Environment, University of Dundee, Nethergate, Dun-dee DD1 4HN, Scotland, 10Department of Anthropology, Laurentian University, Sudbury, Ontario, Canada, P3E 2C6.

Received: 16/3/2013 Accepted: 10/4/2014

Corresponding Author: W.C. Mahaney ([email protected])

ABSTRACT

Both Polybius and Livy described a landslide/landslip that blocked the Punic Army’s exfiltration from a high col on the water divide in the Western Alps. The landslide, more aptly termed rockfall, has been a source of contention amongst classicists for centuries despite the fact that only two cols—Clapier and Traversette—exhibit rockfall debris on the lee side of the Alps. While the Clapier rockfall is too small and too young to have provided blockage, the Traversette debris is nearly as Polybius described it when he re-traced the invasion route some 60 years after the event. His ‘two-tier’ description of the deposit, a doublet of younger and older rock rubble, including measurements of width and volume are close to modern measurements and prove that he knew, in advance, the route Hannibal had followed. It would take a practiced eye to correctly identify the strat-igraphic complexity inherent in the Traversette Rockfall. Here we present weathering ra-tios, soil stratigraphic, mineral, chemical and microbiological evidence in support of Po-lybius’ observations as a considerable background database for future geoarchaeological exploration.

_______________________________________________________________________________ KEYWORDS: Hannibalic invasion, Traversette Rockfall, Physico-mineral-chemical corre-lation to ancient texts. _______________________________________________________________________________

78 MAHANEY et al

© University of the Aegean, 2014, Mediterranean Archaeology & Archaeometry, 14, 2 (2014) 77-94

1. INTRODUCTION The landslide (cf. rockfall; Figs. 1A, 1B)

described by Polybius and Livy (Polybius, trans Scott-Kilvert, 1979; Livy trans. de Sélincourt, 1972) as a major impediment to Hannibal successfully exfiltrating the high Alps during the invasion of 218 BC has been discussed by countless authorities (Dodge, 1891; Wilkinson, 1911; Proctor, 1971; Seibert, 1993; Renaud, 1996; Lazenby, 1998; Bagnall, 1999; Lancel, 1999; Connolly, 1981; Cottrell, 1992; Mahaney, 2004; Ma-haney and Tricart, 2008; Mahaney et al., 2008) over the last two millennia. Few (de Beer, 1967, 1969; Prevas, 1998) have ven-tured forth in search of landslide/rockfall debris on the lee side of the Alps below any of the major cols and fewer still have made any attempt to identify the deposit de-scribed by Polybius (Mahaney, 2008; Ma-haney et al., 2010a).

Of all the environmental targets that would help to identify the correct col of passage by the Punic Army, the search for and identification of a deposit similar to the one described by Polybius would bring to a close the longstanding question of the route Hannibal followed into Italia. The route question, long a source of contention among classicists (Mahaney et al., 2010b), is important more for targeting sites worth geoarchaeological excavation than to satis-fy the musings of armchair theorists, as no artifacts or other geoarchaeological evi-dence of the invasion have ever been found. To confound the issue, even if arti-facts or other evidence were to be found, it would be doubly difficult to pinpoint ma-terial originating from Hannibal’s expedi-tion, as his brother Hasdrubal crossed over into Italia with a large army some eleven years after the main Punic Invasion of 218 BC.

Despite the route controversy that has raged over the post-invasion period, recent critical comment of the Traversette Route (Kuhle and Kuhle, 2012) pinpointed the importance of establishing the stratigraphy

of the two-tier rockfall (Fig. 2) described by Polybius some 60 years after the invasion. As pointed out by Mahaney (2012) the ‘two-tier’ nature of the rockfall mass, origi-nally described by Polybius, is key to iden-tifying the route and presumably a prime locality for geoarchaeological exploration and excavation. The two-tier nature of the rockfall is clearly evident in the field, the older mass of Late Glacial age lying astride the younger mass originating during the Neoglacial.

Polybius’ description of the rockfall: ‘A previous landslide had already carried away some 300 yards of the face of the mountain, while a recent one had made the situation still worse’, (translation by Scott-Kilvert, 1979), sums up the two-tier mass faced by the Punic Army following their descent from the high pass. In Livy’s words the rockfall is described thus (translation by de Sélincourt, 1972): ‘Soon they (Punic Army) found themselves on the edge of a precipice—a narrow cliff falling away…a recent landslide had converted it on this occasion to a perpendicular drop of nearly a thousand feet.’ Livy quotes other sources but never witnessed the mass wasted de-posits at the site as Polybius did, the latter the prime authority on the invasion.

An important proof for the Late Glacial age of the older mass is the presence of cosmic impacted grains in rock rinds at the V9 site (Fig. 2) which have been correlated with the black mat event of 12.8 ka (Ma-haney et al., 2013). The age of the younger debris sheet is based on soil profile charac-teristics that correlate closely with similar Neoglacial-age profiles near Mt. Blanc, ~60 km to the north (Mahaney, 1991). Because the Traversette Rockfall is the only lee-side composite rockfall of two deposits, one su-perposed on the other, it is clearly the mass identified by Polybius during his momen-tous reconstruction of the invasion carried out in the field.

PPOLYBIUS’ ‘PREVIOUS LANDSLIDE’ 79


Figure 1A. Location of sites in the Traversette Rockfall and the Col du Clapier, Italian Alps. The map is from Mahaney et al., 2010.

Figure 1B. Contour map and location of the Traversette Rockfall in the Mt. Viso area.

A search of all major passes, from what

is now the boundary of France and Italy, are of variable height and most trails are steeper on the Italian side as a result of the tectonic structures that have emplaced thrust sheet upon thrust sheet (Mahaney, 2010a). All paths follow steep inclines into Italy, mostly over bedrock, talus and mo-

raine with variable amounts/volumes of rockfall sediment. A search of these areas from south to north for suitable deposits similar to what Polybius described covers all possible routes. Livy, often considered the first modern historian, never left Padua rendering all his descriptions second-hand; whereas Polybius actually interviewed

80 MAHANEY et al


survivors (albeit some sixty years after the invasion when memories were dulled by the passage of time) and followed the inva-sion route across Iberia into southern Gaul (now France) and over the Alps into the headwaters of the Upper Po River (Ma-haney, 2008). Polybius described the high alpine environment providing details of the ‘landslide’ that blocked the army creat-ing an impediment to the passage of the elephant corps, soldiers, horses and mules. The infantry could pass, albeit with diffi-

culty the rubble mass that lay across the slope, but the animals were spooked by the rubble. To find the rockfall that correlates in size and volume with what Polybius de-scribed is to find the route and to prove that Polybius actually visited the site. That is the substance of this report, the search for the great impediment that nearly upset the invasion schedule, finally identified conclusively by Polybius’ description and the stratigraphic evidence reported here.

Figure 2. View across the rockfall looking south. Vegetated areas on both side of the complex deposit approx-

imately comprise the older rockfall member mentioned by Polybius. The fresh unvegetated area is the younger Neoglacial-age deposit.

2. REGIONAL BACKGROUND

The Alps straddle the plate boundary be-tween Europe and Africa which continues to converge today (Tricart, 1984). The Western Alpine arc shared by Italy and France, a broad scale tectonic structure de-veloped along 300 km of longitude with a width of 150 to 200 km, produces a marked asymmetry with the inner eastern slope much steeper than the outer western one. Nearly every researcher who has studied the Hannibal Invasion has focused on this

character of the geological structure, which makes for a difficult descent onto the Po River plains. The orogen asymmetry ex-pressed in the topography originates from the subsurface structural asymmetry of the collision wedge, with repeated outward-directed (west) tectonic transport, the net result of which produced a curvature re-sponding to unequal shortening of the Eu-ropean paleomargin at the western tip of the Adria indenter, an African promontory. The main mountainous relief, oriented NE-



SW in the north and NW-SE in the south, outlines a succession of fold-thrust struc-tures developed deep into the Pre-Mesozoic crystalline basement and Meso-Cenozoic sedimentary cover (Fig.1). Trans-verse valleys across these thrust sheets, through which Hannibal marched to reach the high col, comprise syn-and post-collision faults, but also older faults inher-ited from the initial rift structure.

The core of the Western Alpine arc com-prises a thick pile of metamorphic rock, derived from the subduction of the Tethyan Ocean below its European passive margin, followed by the collision between this margin and the conjugate margin, the Adria margin, on the African side. During this mainly Cenozoic collision, several shortening phases resulted in folds and thrusts, producing the parabolic geometry of the inner mountain belt. From the late Cenozoic onward (Tricart et al., 2006), this polycyclic structure of folds and faults un-derwent gravitational collapse, which re-sulted in transtensional and extensional faulting parallel and transverse to the mountain belt. This late Alpine fault net remains active today and is mainly respon-sible for the location of major valleys, i.e. the Upper Po and Guil catchments. Thus, the geometry of the Alpine fault net pro-vides insight as to the nature of surface ge-omorphological features and why drainage systems appear more complex within the internal arc. It is within the main valley of the Upper Po River that the Traversette Rockfall is located, its origin most likely intricately linked to fault movements fol-lowing deglaciation.

3. METHODS

Rock and soil samples were subjected to various physical, chemical, microbiological and mineralogical analyses as indicated below. Pebbles were collected for lithologic identification and weathering rind analysis to assess relative age following procedures outlined by Mahaney (1990).

Soils were collected and horizons identi-fied following Birkeland (1999); soil colors follow Oyama and Takehara (1970). Soil

samples were subjected to laboratory tests including particle size (Day, 1965) by hy-drometer, conductivity to determine salt content, and pH was determined by elec-trode. The pH and conductivity probes were calibrated before experimentation and the calibration was re-checked at the end and at regular intervals during the measurements. No ‘drift’ occurred. Carbon, nitrogen and hydrogen contents were de-termined using an Exeter Analytical CE 440 elemental analyzer. Soil samples were ground into a fine powder and milligram quantities were analyzed for C, H and N contents. Samples were run in duplicate.

The clay fraction was studied for mineral composition by means of powder XRD us-ing a Bruker 8D diffractometer with Ni-filtered CuKά radiation. Scanning steps for oriented samples were 0.02° 2θ from 2 to 55° 2θ. A semi-quantitative mineral compo-sition was determined from peak integral intensities of chlorite, illite-vermiculite, mi-ca+illite, kaolinite, talc and smectite, multi-plied by factors of 1, 0.35, 2, 1.4, 0.2, 0.2 re-spectively (Whittig, 1965; Kalm et al., 1997).

Sand fractions were imaged by light mi-croscope to determine the composition of the source minerals and their weathered state. These samples were wet washed dur-ing particle size analysis, the sands recov-ered following dry sieving. Approximately 300 grains per sample were examined, and a select few were randomly chosen for mi-croscopic analysis. From this population of grains a subpopulation was selected for more intensive investigation by Scanning Electron Microscope (SEM), Field Emission Scanning Electron Microscope (FESEM) and Energy-Dispersive Spectrometry (EDS) (see-Mahaney, 2002).

Separately collected samples of C/Cox and Cu horizons of the G3, V9 and addi-tional moraine deposits were subjected to preliminary metagenomic microbial analy-sis. Sediment samples were taken under asceptic conditions for this purpose from the mid-point depth of respective soil hori-zons. Cu horizons were sampled approxi-mately 5-cm below the top of the horizon. The samples were immediately stored on

82 MAHANEY et al


ice in the field and frozen at -20 oC upon return to the laboratory. Sediments were subject to total DNA extraction (Griffiths et al., 2000) and subsequently compared us-ing DGGE (Density Gradient Gel Electro-phoresis) analysis of 16S rRNA gene partial sequences (Muyzer et al., 1993). The 16S rRNA gene is present in all bacterial isolat-ed DNA, and as a ‘housekeeping gene’ has been widely used to define microbial phy-logeny. Here populations of this gene in different samples were compared to assess the similarity between bacterial communi-ties found between sites and within soil profiles. Statistical analysis of DGGE re-sults was performed using cluster analysis based upon the unweighted pairwise grouping method with mathematical aver-ages (UPGMA).

4. RESULTS 4.1 The Rockfall

The rockfall (cf. landslide described by Polybius) was said to measure 300 yards, approximately equivalent to 275 m, which is close to the actual size near the source

bedrock where the broad apron of debris measures ~240 m (Fig. 2). Presumably ero-sion over the last 2.2 kyr has converted part of the mass to talus which envelopes the rockfall to the north and south. As shown in Fig. 2 the rockfall exhibits an older vege-tated area with a resident Cryochrept pale-osol of Late Glacial age contrasted against the younger and more voluminous mass of debris carrying a thinner Cryorthent which is largely lacking in significant vegetative cover. Cross sections through the rockfall yield finer debris in the older mass with pockets of large boulders, many with di-ameters of >1 m with similar lichen cover compared to clasts in the larger and younger deposit. However, coarse clastic material in the older deposit displays a rough tank and tor microtopography con-trasting with clasts in the younger deposit which are fresher. In addition to degrees of weathered clast surfaces, pebble rinds at two sites [V8 (Neoglacial) and V9 (Late Glacial)] provide relative age estimations (Table 1).

Table 1. Weathering rind data for sites V8 and V9, Traversette Rockfall, Italian Alps. Rind measurements are in mm. based on populations of 50 at each site. Maximum counts are the Σ of total rind thicknesses at each site. Internal weathering is a summary of the number of clasts carrying internal discoloration in each

population. Site Maximum

count Mean

maximum Minimum

count Mean thickness of minimum

rind Internal

weathering

V8 15 0.30 0 0 4

V9 152 3.04 9 0.18 28 4.2 Weathering Rinds Rock rinds collected and measured at

two sites—V8 and V9—provide useful data on the relative age of the surficial materials at each site. As relative age indicators, clasts lying in residence on deposit surfaces record weathering stresses transmitted to them by the subaerial atmosphere and local biosphere, the chemical front deepening with rinds becoming thicker over time (Mahaney et al., 2013). A 1-mm thick rind means that 1 million nanometers of atomic space has recorded all environmental per-turbations occurring over the lifetime of a

clast resident at the site. Thus, because weathering rind development is differen-tial across a clast (Mahaney et al., 2012), a single rind usually produces a maximum and minimum thickness (Mahaney, 1978). Therefore, a rind population of 50 clasts produces some variability in rind thickness which requires the maximum and mini-mum measurements to be reduced to mean figures. Hence, the values for V8 and V9 as shown in Table 1 record a considerable dif-ference between the Neoglacial and Late Glacial sites. The rind mean maximum thickness at V9 (Late Glacial) is 10 times



the mean maximum rind at V8, a measure of the relative difference between times of emplacement. The mean minimum rinds are different as well with positive values at V9 and nil at V8. This trend is similar to rind measurements taken at other middle latitude sites in North America (Mahaney, 1978) where Neoglacial vs. Late Glacial sites have been differentiated on the basis of mean minimum rind growth.

4.3 The Soils/Paleosol

The soils (Fig. 3) in the younger rockfall member (V2, V4 and V8) and the paleosol in the older member (V9) were analyzed to determine if degrees of difference of nor-mal soil properties might reveal age differ-

ences. The V2 and V8 profiles near the toe of the younger deposit are of similar depth compared with V4 higher up in the depos-it. With thickness running between 22 and 29 cm and Ah/C-Cox/Cu profiles, the soils narrowly permit classification into the Cryorthent (NSSC, 1995) or Regosol (Can-ada Soil Classification, 1977) orders. Color ranges from 10YR 2/1 in the Ah horizons to 10YR 6/1 in the C and 10YR 4/3 in the Cox horizons, the hue in the latter contain-ing sufficient pigment to suggest release of secondary oxides and hydroxides. The Cu horizon color is 2.5Y 5/2, within the nor-mal hue of unweathered parent material.

Figure 3. Soil/paleosol profiles within the rockfall. The soils (V2, V4 and V8) have probably experienced minor climatic perturbations during the later Neoglacial while the Late Glacial paleosol has experienced both a postulated airburst impact (Mahaney and Keiser, 2013) and large scale climatic swings from the

Late Pleistocene to postglacial warming during the Holocene. Field textures remain steady throughout

the pedons as pebbly to coarse pebbly sandy loam in the Cu in V8 where parent material is present. Structure is massive in the lower horizons becoming granular in the Ah group. Moist consistence is massive in the C-Cox/Cu becoming friable in the Ah horizons. All horizons are nonplastic throughout and nonsticky in the C-Cox/Cu and slightly sticky in the Ah suite of sam-ples. Roots extend through into the C hori-zons.

The 22-cm depth of V8 is the thinnest of the three profiles analyzed which indicates the maximum degree of weathering in the younger rockfall member is variable de-

pending upon microenvironment. The oth-er two sites display somewhat thicker pro-files, as well as different approximations of the wetting depth in the surface mantle, i.e. the depth to which soil water penetrates to achieve alteration of fresh mineral material into soil. The change in color in the V8 pro-file follows the same trend as in V2 and V4, exhibiting the standard soil color nomen-clature of Oyama & Takehara (1970), with YR equating to yellow red hues followed by value and chroma making a sharp con-tact with the yellow (2.5 Y) hue of the par-ent material. The surface color is black grading downwards to dull yellowish brown in the C/Cox horizons, the colors

84 MAHANEY et al


indicating a build-up of humus at the sur-face and release of oxides and hydroxides at depth, an arrangement that may well be compatible with the alpine environment for this latitude and elevation, as well as the middle Neoglacial age of the material. Un-dulating horizon contacts shown in Fig. 3 between the Ah and C/Cox horizons sug-gest at least minor freeze thaw activity, not unexpected given the altitude of the site and low clay percentages. The color in the Cox horizon as the ‘ox’ designation implies weak oxidation of mineralic material with colors in the 10YR 4 and 5 range.

Soil structure, which relates to the degree to which fine particles adhere together to form larger aggregates and which takes time to form (Mahaney, 1990), is present only in the Ah horizon in the form of gran-ules, the grade of development similar to what exists in the V2 and V4 profiles de-scribed above.

The V9 profile in the outer and older rockfall member is a Cryochrept (NSSC, 1995) with an Ah/Bw/Cox/Cu profile, of 60-cm thickness and formed over an inter-locking network of coarse clastic rockfall debris. Colors are decidedly stronger, rang-ing from 10YR 2/3 in the Ah and 10YR 4/4 in the Bw horizon, to 10YR 3/4 in the Cox. Field texture is similar to the younger soils with pebbly silty loam in the Ah grading downward to pebbly sandy loam in the Bw and Cox horizons. Soil structure shows an upgrade over the younger profiles with stronger granular clods in the Ah, weak blocky forms in the Bw and massive in the Cox. Moist consistence is similar to that in the younger pedons with a friable condi-tion in the Ah becoming very friable with

depth. Plasticity remains nil, as in the younger profiles, becoming slightly sticky in the Bw and Cox horizons. The Ah/Bw horizon contact shows minor convolutions indicating some frost heave. Roots reach 2 mm in diameter, somewhat larger and more frequent than in the younger soils.

4.4 Particle Size

Particle size (Table 2) was studied in or-der to further assess the effects of age and weathering on the weathered cover sedi-ments in question. Despite a minimum number of samples, the resulting linear particle size curves are characteristic of heterogeneous material produced by large scale mass wasting events, similar to curves generated by till. While clay content in the middle Neoglacial soils is low (~2-3% in all horizons), and somewhat higher in V9 (3.5%), overall the size distributions are not unusual with alpine soils and pale-osols. There is no demonstrable downward movement of fines within the profiles. The humus-rich Ah horizons are distinct from the C/Cox horizons in all profiles in that there is approximately higher silt in the subsurface than at the surface, which ar-gues against aeolian influxes of fine grained material. Calculated mean phi val-ues, determined on the 25th, 50th and 75th percentiles support the grade size results reported for each silt grade size. The cen-ter-of-gravity measurements of 2.3 to 3.5 mean ø indicate only minor downward shifts in clay which may be the result of high porosity or settling-out processes in coarse grained rockfall deposits rather than weathering over time.

Table 2. Particle size distributions in soils and paleosols of the Traversette Rockfall.

Site Horizon Depth (cm) Sand (2mm-63µm) (%)

Silt (63-2 µm) (%)

Clay (<2 µm) (%)

Mean Phia

Viso2 Ah 0-15 70.9 27.1 2.0 2.3

Cox 15-35 64.9 33.4 1.7 3.0 D 35+ -- -- -- --

Viso4 Ah 0-9 69.6 29.1 1.3 2.8 Cox 9-27 60.5 37.2 2.3 3.0 D 27+ -- -- -- --



Viso8 Ah 0-10 63.8 35.2 1.0 2.9 C 10-22 63.7 34.7 1.6 3.0 Cu 22+ 63.7 34.5 1.8 3.0

Viso9 Ah 0-12 64.5 32.5 3.0 3.1 Bw 12-22 54.9 44.1 1.0 3.5 Cox 22-60 52.9 44.1 3.0 3.5 D 60+ -- -- -- --

a Mean phi is calculated: Σ= 25th + 50th + 75th %/3.

4.5 Clay Mineralogy Distinctions between the Ah and C hori-

zons of the young member profiles (V2, V4 and V8) are minor as indicated in Fig. 4. Similar concentrations of illite in both hori-zons of V4 and V8 indicate no chloritiza-tion of illite, as seen in other Neoglacial profiles (Mahaney, 1978), but rather inher-itance from the local mica schist bedrock. The presence of illite-vermiculite with greater concentrations in the C horizon of V4 suggests a possible link with boreal for-est species which may have inhabited the site in the Atlantic and sub-Boreal

chronozones (see pollen section below). The small concentrations of amphibole (hornblende) may result from airfall influx or the local source rock. Minor pedogenic modification of the mineral composition of the soils supports a young mid-Neoglacial age for profiles V2 - V8. The V9 profile is similar to V4 and V8 but with slightly greater chlorite which could relate to chloritization of illite. Quartz and plagio-clase are variable among the three profiles and in trace quantity only but overall do not allow discrimination between sites.

Figure 4. XRD traces across various horizons in the younger rockfall member soils and the V9 paleosol. 4.6 Soil Chemistry

86 MAHANEY et al


Soil chemical properties of profile V8 (Neoglacial) and V9 (Late Glacial) (Table 3) provide a database on the compositional differences between deposits in the rock-fall. Aside from the taxonomic differ-ences—Cryorthent (V8) to Cryocrept (V9)—pH is strongly acid in both profiles with only very slightly greater pH in the younger system. Down profile in both sys-tems pH decreases with increased leaching time becoming somewhat more acidic in the older V9 profile. Electrical conductivity, as a measure of total salt, is variable in the two systems. In V8 total salt decreases with depth into the C horizon, increasing in the parent material presumably sourced from the bedrock. The distribution in V9 depicts a slow decrease with depth with overall lower amounts possibly produced by in-creased leaching in the older profile. Or-ganic carbon in both profile epipedons (Ah horizons) yield similar concentrations sug-gesting that C/N dynamic equilibrium is reached in less than ~3 ka and possibly sooner. Downward trends of C/N in both profiles are similar.

4.7 Microbiology

A comparison of the microbiological eu-bacteria populations was performed using DGGE analysis of the 16S rRNA genes in a number of C/Cox and Cu horizons – in-cluding the V8 and V9 C/Cox horizons.

Soil microbiological population structures are complex, and the individual compo-nents of a population are influenced by a wide range of factors including the imme-diate microclimate chemical/physical components in the sediment matrix and the age/history of the sediment. With this mul-titude of factors in mind a DGGE (density gradient gel electrophoresis) comparison was used to try and show if the younger V8 C/Cox population differed to any extent from the apparently older V9 C/Cox, other C/Cox sediments in the area, and the asso-ciated Cu sediments. In Fig. 5 a dendro-gram shows the close relation of the eubac-terial populations in these sediments. Two key factors are apparent: i) That the C/Cox vs. Cu horizons have distinctly different populations; ii) That the Viso 8 population is less similar to the other C/Cox popula-tions studied in this experiment; iii) that the C horizon in V8 is distinct from V9, the latter showing a close correlation with oth-er C horizons of Late Glacial age (15-13 ka) in soil profiles of the Viso area. Point (ii) is consistent with a view that V8 represents younger sediment – perhaps characterized by less weathering and therefore differing geochemistry in the microenvironment, or – in the absence of a Bw horizon – greater exposure to micronutrients percolating down from the Ah horizon.

Table 3. Selected chemical parameters for soil (V8) and paleosol (V9) in the Traversette Rockfall. Site Horizon Depth (cm) pH (1:5) Conductivity

µS/cm-1 Organic C (%) TKN (%)

V8 Ah 0-10 5.6 56.0 6.49 0.49 C 10-22 5.2 19.3 1.60 0.04 Cu 22+ 5.6 116.8 0.69 0.00

V9 Ah 0-12 5.4 41.0 6.79 1.01 Bw 12-22 4.9 26.1 1.89 0.07 Cox 22-60 5.1 4.9 1.30 0.06 D 60+ -- -- -- --

4.8 Scanning Electron Microscopy

Randomly selected mineral grains from the various horizons of profile V4 and V9 were subjected to analysis by SEM-EDS to assess their weathered state. The results for V4, previously published by Mahaney et al.

(2010a) indicate that over 80% of the grains contain dense accumulations of adhering particles, appearing indurated on the min-eral surfaces (Mahaney et al. 2010a; Figs. 9A and B). There does not appear to be any



cementation of grains or melted grains in the younger profile, but none is expected given the young age of the V4 soil. Most grains exhibit minor subparallel fractures and abrasion features as well as fracture faces, the latter probably resulting from bedrock release by physical weathering. These microfeatures are strikingly similar to microtextures found in fine talus debris in other mountainous areas (Mahaney, 2002) and are quite distinct from microtex-tures related to other geological processes.

The indurated nature and lack of cemen-tation of the matrix material in the younger rockfall lobe implies that Hannibal’s engi-neers could segregate large clasts to pro-vide a wide pathway through the rubble. Removing the clasts by levering them out position was probably difficult given the physical state of the infantry and the over-all decline of morale.

Microtextural analysis of the sand frac-tions in the older rockfall lobes reveals a multitude of grains (Fig. 6A) with adhering particles fused together from a high tem-perature event, presumably the black mat impact of 12.8 ka as documented nearby in the upper Guil River Catchment (Mahaney and Keiser, 2013). The EDS spectrum shown in Fig. 6B depicts the common melt-ed and fused nature of small carbon and Fe spherules welded to a very fine sand py-roxene grain.

4.9 Rockfall Reconstruction: Late Glacial to Mid Neoglacial age.

Lichen growth (total cover and maxi-mum diameters of Rhizocarpon section ge-ographicum) are close to maximum, and hence, near longevity on rock outcrops above the rockfall, which means that a li-chen trimline that might delineate a younger bedrock source area for the Neo-glacial debris does not exist.

Figure 5. Microbiology analysis consisting of a DGGE comparison of the eubacterial 16S rRNA gene pop-ulations from two Cu horizons, and 4 C/Cox horizons – including V8 and V9—of a suite of Late Glacial paleosols and soils. The data can be interpreted to show how, relatively, closely related the populations are in these 8 samples. The pp G7 negative control represents a single laboratory Pseudomonas putida

strain (a common soil bacterium) for comparison.

88 MAHANEY et al


Figure 6. A, SEM image of a pyroxene of very fine sand size in the outer and older lobe of the Traversette

rockfall. The sample contains a multitude of welded pyroxene, quartz and Fe spherules, mostly coated with carbon from wildfire that engulfed the site following a cosmic airburst at 12.8 ka; The grain is highly

fractured and covered with adhering particles presumably from the mass wasted event.

Figure 6B. EDS of a welded grain showing a typical augite signature (Ca unmarked) along with very high

carbon. The small concentration of Na may be related to acmite fused with augite.

Moreover, maximum diameters of R. ge-ographicum on the bedrock outcrop reach 95 mm, similar to maximum diameters on both the younger and older rockfall debris members. Lichen cover, somewhat higher on the rockfall material than on the bed-rock walls, ranges from 60 to 75%, which is at or close to longevity at the nearest locale (i.e. Mt. Blanc, Mahaney, 1991). The varia-tion in lichen cover, with lower values on bedrock walls, is most probably due to as-

pect, wind direction, prevalent ice crystal blasting and moisture variations. In places, numerous plumes of Xanthoria elegans lace bedrock and rockfall surfaces where bird perches exist, associated lichen fertilized by various avian species, as documented elsewhere (Julig et al., 2012). Whilst the fer-tilized colonies of Zanthoria show inordi-nately high percent cover this does not seem to have affected the growth and spread of Rhizocarpon.



Moreover, a close scrutiny of rock out-crops above the rockfall does reveal a weathering trimline; heavy oxide coatings above on the recessed ridge slope 30 m above the upper deposit and lighter tonal contrast below suggests the younger age probably outlines the source area for sedi-ment in the younger rockfall member. While weathering characteristics allow rel-ative age discrimination, lichen cover and maximum lichen diameters reveal that both rockfall deposits are older than lichen lon-gevity, both on the exposed bedrock wall above the upper rockfall debris and in the general area (Pech, et al., 2003). The lichen cover by itself argues for an age >2 ka (Ma-haney et al., 2010a). 4.10 Paleo-Vegetation Reconstruction

The vegetation proxies from small lakes reveal that the subalpine belt was mostly covered by woodland until the Middle Age (1800-1300 cal yr BP) depending on sites. The upper timberline, dominated by cem-bra pine, occupied a position some 300 and 500 m above the present-day timberline before 2000 cal yr BP, although it had al-ready begun to shift downslope approxi-mately 4000 cal yr BP (Carcaillet et al. 1998; Ali et al. 2005; Carcaillet and Muller 2005). Using palynological data as a guide it is probable forest reached to an elevation of ~2300 m or just below the rockfall. Recov-ered core data indicate that with sufficient biomass, subalpine wildfires occurred in-frequently at a rate of 3-5 fires/1000 years. The conversion to subalpine meadow and short grassland occurred after 2000-1800 cal yr BP producing a progressive decrease of tree-pollen frequency and abundance (Genries et al. 2009) associated with fires resulting from fuel-suppression (Carcaillet et al. 2009) resulting from a change of land-use occupied by cattle and sheep or from onset of the sub-Atlantic chron in the late Neoglacial. Both plant macro-remains from several sites and pollen data (e.g. David 1995; Genries et al. 2009) demonstrate that forest was the dominant community until the widespread distribution of grass be-tween 2000-1300 cal yr BP. When Hannibal and his army passed the Traversette Rock-

fall, scattered trees were likely present just below the rubble sheet. The climate was more or less similar to today.

5. DISCUSSION

These data indicate that deposit mor-phology clearly shows a major difference between the two members of the Traversette Rockfall, an older deposit deep-ly buried by a second mass of somewhat greater volume. When Polybius visited the site the younger deposit may have had a steeper slope than today with larger clasts near the bedrock wall. The deposit may also have had a narrow path extending to the north which would have had to be widened, ballast fixed in place to allow the passage of animals. To the starving troops this must have seemed a daunting task. Scott-Kilvert’s (1979) translation of Polybi-us makes this point crystal clear: ‘He (Han-nibal) set his troops to work on the im-mensely laborious task of building up the path along the cliff. However in one day he had made a track wide enough to take the mule train and the horses; he at once took these across, pitched camp below the snow-line and sent the animals out in search of pasture. Then he took the Numid-ians and set them in relays to the work of building up the path. After three days of this toilsome effort he succeeded in getting his elephants across, but the animals were in a miserable condition from hunger. The crests of the Alps and the parts near the tops of the passes are completely treeless and bare of vegetation, because of the snow which lies there continually between win-ter and summer, but the slopes half-way down on the Italian side are both grassy and well-wooded, and are in general quite habitable.

No doubt the present trail extending from an old Italian Army barracks higher up the mountain to the switchback leading into the lower valley has been repaired and re-ballasted several times, mainly with rock taken out of the younger rockfall member. Taking Polybius’ account at face value it is difficult to understand the recent criticism of Mahaney et al. (2010a) by Kuhle and

90 MAHANEY et al


Kuhle (2012) who argue Hannibal cut a path across ledges there being no landslide when it is clear from Polybius’ account that the landslide was the obstacle. Otherwise why would he have identified the landslide and ‘building up a path’ which surely re-fers not to carving out a ledge route but shoring up a path through the landslide, cf. rockfall. The illogic of such a criticism is reinforced all the more by reference to the Col du Clapier (Fig. 1A), 60 km north of the Traversette, considered by Kuhle and Kuhle (2012) as the favored route of pas-sage of the Punic Army, an entrance point to Italia that does not match the environ-mental description in any of the ancient texts (Mahaney, 2008).

The variation from a clast-dominated matrix or open network of coarse material below (D horizon) to a more dispersed ar-rangement of smaller stones mixed with sand and finer material above suggests a special microenvironment where plants have stabilized the subsurface. For the most part the soils, even the one paleosol, described here do not contain Cu horizons of fine grade size material. Soil thickness and horizon designation in the younger member group of soils depends upon mi-croenvironment and position in the depos-it. The V2 profile in the upper rockfall formed over an open network of coarse clastic sediment, presumably supported and stabilized by plant roots. Lower in the rockfall the V4 and V8 profiles carry simi-lar profiles but with occasional collection of silty sand containing a Cu horizon making it possible to study the transition of parent material into soil. This transition clearly shows little change in particle size, the main weathering products confined to col-or changes marking the production of ox-ides and hydroxides. Contrasting these young member soils with the older paleo-sol at V9 is highlighted by the production of a fully formed Cryochrept soil of twice the depth and somewhat stronger color in-dicative of greater chemical weathering but with similar particle size distributions, per-cent clay content and clay mineral compo-

sition. Otherwise there is little to discrimi-nate between the two groups.

A search of recovered soil for telltale flakes of charcoal, evidence that might support Livy’s contention that Hannibal fired rock on the lee side of the Alps, proved unsuccessful. As discussed, the on-ly route with fired bedrock outcrops lies through the Col du Clapier, 50 km to the north of the Col de la Traversette (Sodhi, et al, 2006; Mahaney et al., 2007; Mahaney, 2008).

Even though soil stratigraphy—sequences of soils used to date underlying deposits---in the Western Alps has never been established (see Egli, et al., 2003) it is possible, in a general way, to correlate the younger soil in the Traversette rockfall with places further afield, where approxi-mate ages of soils are known (Mahaney, 1991). Using depth of weathering, A/C soil horizon assemblage, soil colors and in-ferred organic and weathering compo-nents, textural gradations down-profile, and low grade of soil structure it is likely these profiles carry a Neoglacial age, most probably between 2000 and 4000 years. While it is difficult to tie these profiles di-rectly to Hannibal, it is likely he crossed over the younger profiles on the march into Italia.

The older paleosol (V9), is the one profile selected as representative of the profile ex-pression in the older rockfall member. As with the younger soils, pH in the V9 profile is strongly acidic, and contains somewhat greater H+ ion content in the Bw horizon. The total salt content decreases down pro-file from higher concentrations in the Ah horizons where nitrates are released by vascular plants to lesser amounts released by leaching. Similarity of pH in all surface horizons indicates that soils are at or close to dynamic equilibrium at all sites in the younger rockfall member.

6. CONCLUSIONS

The 2200-yr old description by Polybius of the rockfall that blocked Hannibal and the Punic Army on their high-col crossover to Italia during the great invasion has been



the subject of debate by classicists for cen-turies. Polybius’ first-hand account depicts a two-tier mass wasted deposit of spectacu-lar size and scope, one sufficient to block cavalry horses, draft animals and elephants requiring some three days for his engineers to clear, ballast and forge a path through to a regrouping area in the lower valley. Livy, echoing Polybius a century and half later and relying on other sources, described a single mass wasted deposit blocking the Carthaginians, and noted it had a width of 300 yards. No doubt Polybius’s description is preferred over Livy as the great general actually followed the invasion route and was presumably on site describing the ‘landslide’ in great detail.

Whilst little attention has been paid to the size and scope of landslides or rockfalls on the exit routes of the major passes, the identification of the Traversette Rockfall as the only sizable mass wasted deposit matching Polybius’s description, supports de Beer (1969) who, for a variety of reasons, contended that Hannibal crossed the high-

est of cols in the Western Alps to gain en-trance to Italia. The stratigraphic data pre-sented here not only supports de Beer but also lays out the geologic database that clearly shows the rockfall size and scope as well as differentiation of the two rockfall members, the oldest assigned to the Late Glacial stage of the Last Glacial Maximum and the youngest to the middle Neoglacial. What is most crucial here is that for the first time an area comprising some 240 m x 20 m is fertile ground for geoarchaeological exploration. It is impossible to imagine an army the size of Hannibal’s force, one en-gaged in clearing and ballasting a path through a rubble sheet, not losing or dis-carding implements and personal accou-trements. As a prime target for such explo-ration, on a similar scale to the bivouac in the Upper Guil River to the west of the high col and the regrouping area of the Upper Po River, the rockfall is a prime area for a ground penetrating radar / geophysi-cal metal detection survey.

ACKNOWLEDGEMENTS

This work was funded by Quaternary Surveys, Toronto. REFERENCES Ali, A.A., Carcaillet, C., Talon, B., Roiron, P. and Terral, J.-F. (2005) Pinus cembra (arolla), a

common tree in the inner French Alps since the early Holocene and above the present tree line: a synthesis based on charcoal data from soils and travertines. Journal of Biogeography, vol. 32, 1659-1669.

Bagnall, N. (1999) The Punic Wars. Pimlico, London, 347 pp. Birkeland, P.W. (1999) Soils and Geomorphology, Oxford University Press, Oxford, U.K.,

430 pp. Canada Soil Survey Committee (1977) Soils of Canada 1. Agriculture Canada, Ottawa. 243

pp. Carcaillet C. and Muller S.D. (2005) Holocene tree-limit and distribution of Abies alba in

the inner French Alps: anthropological or climatic changes? Boreas, vol. 34(4), 468-476.

Carcaillet, C., Talon, and B., Barbero, M. (1998) Pinus cembra et incendies au cours de l'Holocène, 300 m au-dessus de la limite actuelle des arbres dans les Alpes du nord-ouest. Écologie, 29, 277-282.

Carcaillet, C., Ali, A.A., Blarquez, O., Genries, A., Mourier, B. & Bremond, L. (2009) Spa-tial variability of fire history in subalpine forests: large and local scale processes during the Holocene (from natural to cultural regimes). Ecoscience, 16(1), 1-12.

Connolly, P. (1981) Greece and Rome at War. MacDonald and Co., Cottrell, L. (1992) Hannibal Enemy of Rome. Da Capo Press, N.Y., 257 pp.

92 MAHANEY et al


David, F. (1995) Vegetation dynamics in the northern French Alps. Historical Biology, vol. 9, 269-295.

Day, P. (1965) Particle fractionation and particle size analysis, In, C.A. Black, ed., Methods of Soil Analysis, Madison, Wi., Am. Soc. Agronomy, 545-567.

de Beer, G. (1967) Hannibal’s March, Sidgwick and Jackson, London, 164 pp. de Beer, G. (1969) Hannibal. The Viking Press, N.Y., 320 pp. Dodge, T.H. (1891) Hannibal, Houghton Mifflin, N.Y., Egli, M., Mirabella, A., Sartori, G. and Fitze, P. (2003) Weathering rates as a function of

climate: results from a climosequence of the Val Genova (Trentino), Italian Alps, Geoderma, vol. 111, 99-121.

Genries A., Muller S.D., Carcaillet C., Mercier L. & Bircker L. (2009) Local-scale processes control spatial variability of subalpine vegetation trajectories during the Holo-cene. Ecoscience, vol. 16 (1), 13-22.

Griffiths, R.I., Whiteley, A.S., O’Donnell, A.G., and Bailey, M.J. (2000) Rapid Method for Coextraction of DNA and RNA from Natural Environments for Analysis of Ribo-somal DNA- and rRNA-Based Microbial Community Composition. Applied and Environmental Microbiology, 66(12), 5488-5491.

Julig, P., Mahaney, W.C., Kalm, V., Earl-Goulet, J. (2012) Geoarchaeological importance of Subarctic bird perches dated by turf mound accumulation and identified by lichen plumes. Polar Record, doi:10.1017/S0032247412000368.

Kalm, V., Kriiska, A. and Aruväli, J. (1997) Mineralogical analysis applied in provenance studies of Estonian Neolithic pottery. — Eesti TA Toimetised. Geoloogia, vol. 1, 16–34.

Kuhle, M., and Kuhle, S. (2012) Hannibal gone astray? A critical comment on W.C. Ma-haney et al., ‘the Traversetted (Italia) rockfall; geomorphological indicator of the Hannibalic invasion route, Archaeometry, vol. 52 (1), 156-172.

Lancel, S. (1999) Hannibal, Blackwell, Oxford, 243 pp. Lazenby, J.F. (1998) Hannibal’s War, University of Oklahoma Press, Norman, 340 pp. Livy, trans. Aubrey de Sélincourt (1972), The War with Hannibal, Penguin, London, U.K,

711 pp. Mahaney, W.C. (1978) Late Quaternary soil stratigraphy in the Wind River Mountains,

western Wyoming, In. W.C. Mahaney ed., Quaternary Soils, Geo-Abstracts, Nor-wich, UK, 223-264.

Mahaney, W.C. (1990) Ice on the Equator, Wm. Caxton Ltd., Ellison Bay, Wisc., 386 pp. Mahaney, W.C. (1991) Later Pleistocene and Holocene glacial chronology at Chamonix

and Argentière, French Alps. Zeitschrift für Geomorphologie, vol. 35, 225-237. Mahaney, W.C. (2002) Atlas of sand grain surface textures and applications. Oxford Universi-

ty Press, Oxford, UK., 237 pp. Mahaney, W.C. (2004) A geological/topographical reconnaissance of Hannibal’s invasion

route into Italia in 218 BC. In, D.R. Caldwell, J. Ehlen and R.S. Harmon, eds., Stud-ies in military geography and geology, Kluwer Academic Publishers, Dordrecht, Hol-land, p. 67-78.

Mahaney, W.C. (2008) Hannibal’s Odyssey: environmental background to the alpine invasion of Italia. Gorgias Press, Piscataway, N.J., 345 pp.

Mahaney, W.C. (2012) Comments on Kuhle & Kuhle, Archaeometry, ‘Hannibal gone astray’ a critical comment on the Hannibalic invasion route. Archaeometry, doi:10/j.1111/j.1475-4754.2012.00724.x.

Mahaney, W.C. and Tricart, Pierre (2008) Hannibal’s debacle in the Combe de Queyras in 218 BC: the unknown Gallic Commander, In Military Geography and Geology: Histo-



ry and Technology, C. P. Nathanail, R.G. Abrahart and R.P. Bradshaw, eds. ISBN number 095477429. Land Quality Management Press, Nottingham, U.K., p. 88-97.

Mahaney, W.C., Keiser, L. (2013) Weathering Rinds: Unlikely Host Clasts for Evidence of an Impact-Induced Event. Geomorphology, vol. 184, 74-83.

Mahaney, W.C., Milner, M.W., Sodhi, Rana, Dorn R.I., Boccia, S., Beukens, R.P., Tricart, Pierre, Schwartz, Stephane, Chamorro-Perrez, E., Barendregt, R.W., Kalm, Volli, Dirszowsky, Randy W. (2007) Analysis of burnt schist outcrops in the Alps: rela-tion to historical archaeology and Hannibal’s crossing. Geoarchaeology, vol. 22, 797-816.

Mahaney, W.C., Kalm, V., and Dirszowsky, R. (2008) The Hannibalic Invasion of Italia in 218 BC: geological/topographic analysis of the invasion routes, In Military Geog-raphy and Geology: History and Technology, C. P. Nathanail, R.G. Abrahart and R.P. Bradshaw, eds. Land Quality Management Press, Nottingham University, Not-tingham, U.K., p. 76-86.

Mahaney, W.C., Kapran, B., Kalm, V., Tricart, P., Carcaillet, C., Barendregt, R.W., and Somelar, P. (2010a) The Traversette rockfall: geomorphological reconstruction and importance in interpreting classical history. Archaeometry, vol. 52, 156-172.

Mahaney, W.C., Tricart, P., Carcaillet, C., Blarquez, O, Ali, A.A., P., Agnant, J., Bar-endregt, R.W., Kalm, V. (2010b) Hannibal’s invasion route: age-old question revis-ited with new scientific data. Archaeometry, vol. 52, 1096-1109.

Mahaney, W.C., Krinsley, D.H., Allen, C.C.R., Langworthy, K., Ditto, J., and Milner, M.W., Weathering rinds: archives of paleoenvironments on Mount Kenya, East Africa (2012), Journal of Geology, vol. 120, no. 6, 591-602.

Mahaney, W.C., Keiser, L., Krinsley, D.H., Allen, C.C.R., Dirszowsky, R., and Julig, P. (2013). Weathering rinds as mirror images of palaeosols: examples from the West-ern Alps with correlation to Antarctica and Mars Journal Geological Society, vol. 170, 838-847.

Muyzer, G., de Waal, E.C., and Uitterlinden, A.G. (1993) Profiling of complex microbial populations by denaturing gradient gel electrophoresis analysis of polymerase chain reaction-amplified genes coding for 16S rRNA. Applied and Environmental Microbiology, vol. 59(3), 695-700.

National Soil Survey Center (NSSC) (1995) Soil Survey Laboratory Information Manual. Soil Survey Investigations Report No. 45. Version 1.00. USDA, Washington, DC., 305 pp.

Oyama, M. and Takehara, H. (1970) Standard Soil Color Charts, Japan Research Council for Agriculture and Fisheries.

Pech, P., Jomelli, V., Baumgart-Kotarba, M., Bravard, J-P., Chardon, M., Jacob, N., Kedzia, S., Kotarba, A., Raczkowska, Z. and Tsao, C. (2003) A lichenometric growth curve in the French Alps: Ailefroide and Veneon valleys; Massifs des Ecrins. Geo-dinamica Acta, vol. 16, 187-193.

Polybius, trans. Ian Scott-Kilvert (1979) The Rise of the Roman Empire, Penguin, London, U.K., 574 pp.

Prevas, J. (1998) Hannibal Crosses the Alps, Sarpedon, Rockville Centre, N.Y., 232 pp. Proctor, Dennis (1971) Hannibal’s March in History, Oxford University Press, Oxford, U.K.

229 pp. Renaud, J.P. (1996) L’itineraire transalpine d’Hannibal. La fin d’une enigma, Archeologia.

Prehistoire et Archeologie-Dijon, vol. 324, 48-55. Seibert, J. (1993) Hannibal, Wissenschaftliche Buchgesellschaft, Darmstadt, 75-134. Sodhi, R., Mahaney, W.C., and Milner, M.W. (2006) ToF-SIMS applied to historical ar-

chaeology in the Alps, Applied Surface Science, vol. 252, 7140-7143.

94 MAHANEY et al


Tricart, P. (1984) From passive margin to continental collision: a tectonic scenario for the Western Alps, American Journal of Science, vol. 384, 97-120.

Tricart, P., Lardeaux, J.-M., Schwartz, S., and Sue, C. (2006) The late extension in the inner Western Alps: a synthesis along the south-Pelvoux transect. Bulletin de la Société Geologique de France, vol. 177, 299-310.

Whittig, L.D. (1965) X-ray diffraction techniques for mineral identification and mineralog-ical composition. In Black, C.A., ed., Methods of Soil Analysis, V. 1, Madison, Wisc., Amer. Soc. Agronomy, 671-696.

Wilkinson, H. Spencer (1911) Hannibal’s March Through the Alps. Oxford at the Clarendon Press (in a digitized form by internet archive with funding from Microsoft corpo-ration 2007)

Documents

POLYBIUS''PREVIOUS LANDSLIDE': PROOF THAT HANNIBAL'S INVASION ROUTE CROSSED THE COL DE LA TRAVERSETTE